Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Renewable and Sustainable Energy Reviews 133 (2020) 110311

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews


journal homepage: http://www.elsevier.com/locate/rser

Review of the cooperation and operation of microgrid clusters


F. Bandeiras a, E. Pinheiro a, M. Gomes a, b, *, P. Coelho a, b, J. Fernandes a
a
Instituto Politécnico de Tomar, Tomar, Portugal
b
Smart Cities Research Center (Ci2-IPT), Tomar, Portugal

A R T I C L E I N F O A B S T R A C T

Keywords: Multiple microgrids can operate when interconnected and form a cluster of microgrids, in which each individual
Microgrid cluster system benefits from this cooperation during grid-connected and islanded modes. Therefore, the contents of this
Control paper address the concept of microgrid clusters by providing a review of the literature research conducted to­
Energy management
wards the project and development of smart grids. Several aspects regarding the operation of microgrid clusters
Electrical protection schemes
Energy trading
are introduced, including control and energy-management strategies and architecture configurations in terms of
Internal energy market layout, power conversion technology and line frequency technology. A brief comparison of these control stra­
tegies and architecture configurations is also provided. In the field of electrical protections, most of the electrical
protection schemes being proposed in the literature are applied exclusively to individual microgrid systems. Even
though only a handful of works and studies focus on the protection of multiple microgrids, this paper contributes
a review of electrical protection schemes currently available in the literature for microgrid clusters. In addition, it
also addresses energy trading among multiple microgrids and introduces three energy-market designs suitable for
implementation in microgrid clusters to facilitate energy trading among the participating prosumers, producers
and customers. Finally, a case study is presented to evaluate the cooperation among five industrial microgrids
operating in a cluster during islanded mode using an internal market. Each microgrid participating in the
network can sell or buy excess energy in order to fulfil its own power requirements. To this end, an algorithm was
developed in Matlab, allowing the coordination of the hourly energy trade among these microgrids through three
market models with market clearing price/quantity in asymmetric pool.

initially emerged not only as a way of reducing greenhouse gas (GHG)


1. Introduction emissions associated with the use of fossil fuels, but also as a way of
solving the power supply crisis in isolated and remote locations with
A microgrid (MG) is a small-scale electrical power grid which con­ defective and unreliable distribution grids and of guaranteeing the
sists of microgeneration units, storage units and controllable loads. MGs reliability of the power supply in critical facilities with high security
are intended to ensure efficient energy management by coordinating the requirements (e.g. airports, hospitals, data centres, military in­
available energy resources at their disposal. This can only be success­ frastructures) [4].
fully achieved by taking economic, technical and environmental con­ The widespread implementation of MGs can contribute a significant
cerns into consideration. Moreover, several control and protection portion of the global energy required and eventually replace conven­
devices ensure that the MG is capable of handling both grid-connected tional natural gas and coal-fired power plants. In the first quarter of
and islanded operation modes in the safest, most optimal way possible 2020, Guidehouse Insights identified 6610 MG projects which have been
[1]. During grid-connected operation, the MG is connected to the public proposed, planned and deployed around the world, representing a
grid at the point of common coupling (PCC), enabling active interaction combined installed power capacity of around 31.8 GW [5]. Even though
with the distribution system operator (DSO). During islanded operation, the number of MG systems has been rapidly increasing, it is not yet
the MG is disconnected from the public grid and manages the available nearly large enough to completely cover the energy supplied by con­
resources in order to ensure a continuous supply of power to the critical ventional power generation systems. However, it cannot be denied that
loads. MG systems can, in fact, increase the quality and reliability of the MGs have been successful at addressing the growing energy demand and
power supplied to the loads, and mitigate the environmental impact environmental issues and at providing several technical and economic
from the use of fossil fuels [2,3]. The project and development of MGs benefits. Industrial and commercial buildings can effectively benefit

* Corresponding author. Instituto Politécnico de Tomar, Tomar, Portugal


E-mail address: mgomes@ipt.pt (M. Gomes).

https://doi.org/10.1016/j.rser.2020.110311
Received 10 March 2020; Received in revised form 19 August 2020; Accepted 24 August 2020
Available online 2 September 2020
1364-0321/© 2020 Elsevier Ltd. All rights reserved.
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

List of abbreviations IGBT Insulated-Gate Bipolar Transistor


IPFC Interline Power Flow Controller
AC Alternating Current LC Load Controller
ADMM Alternating Direction Method of Multipliers MC Microsource Controller
BTB Back-to-Back MG Microgrid
CAMC Central Autonomous Management Controller MGCC Microgrid Central Controller
CB Circuit Breaker MO Market Operator
CCG Canonical Coalition Game P2G Prosumer-to-Grid
DC Direct Current P2P Peer-to-Peer
DER Distributed Energy Resources PCC Point of Common Coupling
DG Distributed Generation PLC Power-Line Communications
DMS Distribution Management System PoA Proof-of-Authority
DSL Digital Subscriber Lines PoS Proof-of-Stake
DSO Distribution System Operator PoW Proof-of-Work
EMPC Distributed Economic Predictive Control PV Photovoltaic
ESS Energy Storage System QoS Quality-of-Service
FIS Fuzzy Inference System RES Renewable Energy Sources
GHG Greenhouse Gas SOCP Second-Order Cone Programming
HIF High Impedance Fault SSCB Solid-State Circuit Breaker
HUCC Hybrid Unit of Common Coupling WLS Weighted Least Square
ICT Information and Communication Technology WMN Wireless Mesh Network
IEC International Electrotechnical Commission WT Wind Turbine
IEEE Institute of Electrical and Electronics Engineers

from MG systems, especially to mitigate potential voltage drops in strategies, architecture configurations and energy-trading schemes, as
quarries, mines and other extractive enterprises which are usually well as electrical protection schemes currently available in the literature
located far away from population centres, or to prevent significant for MG clusters.
financial losses in commercial buildings such as hypermarkets and This paper is organized as follows: Section 2 introduces the concept
shopping malls which require a robust power supply. An MG owned by of MG clusters and recent trends in their development. Section 3 con­
industrial and commercial organizations will mainly focus on economic centrates on the control and energy management of interconnected MGs
targets and the reliability of the power supply, while an MG owned by a by providing a comparison of different types of control architectures and
government or military entity is likely to focus more on energy security a review of the most relevant works carried out based on each control
and safety. Utility-owned MGs aim at serving customers with special architecture. Section 4 addresses protection challenges associated with
local needs and at ensuring service quality along the distribution grid the operation of MGs and presents a brief comparison of electrical
[6]. protection schemes proposed for MG clusters. Section 5 focuses on the
The traditional way to produce and use energy has been transformed architecture of MG clusters in terms of layout, power conversion tech­
into a share of energy due to the increased penetration of distributed nology and line frequency technology. Section 6 addresses energy
energy resources (DER) into the power systems. This share of energy has trading among multiple MGs and introduces energy-market designs
turned passive consumers into active prosumers responsible for sharing suitable for implementation in MG clusters. Section 7 presents a case
their excess electricity generated by renewable energy sources (RES) study to evaluate the cooperation of five industrial MGs and coordinate
with the grid or other consumers [7,8]. Current energy markets and their hourly energy trade in islanded mode through three distinct market
conventional distribution grids may not be fully prepared to deal with a models with market clearing price/quantity in asymmetric pool. Finally,
more decentralized share of energy. The implementation of smart grids section 8 presents the main conclusions.
and MG systems can facilitate this energy trading among prosumers
through new and reorganized energy-market designs. 2. Clusters of MGs
Therefore, this paper contributes to the project and development of
MG clusters and smart grids by providing a review of the most relevant The concept of MG can be extended to a larger scale, in which two or
works presented in the literature. The concept of MG clusters has been more MGs operate interconnected with each other to form a cluster of
extensively discussed in the literature, focusing mainly on control and controllable MGs. The interconnection of multiple MGs can be recog­
energy management of networked MGs and energy-market designs for nized as a viable solution to mitigate the unstable operation of indi­
energy trading among MG systems. In addition, some of the works vidual MGs during islanded operation and provide numerous economic
presented in the literature address the architecture and layout of MG benefits to both the utility grid and the MG systems participating in the
clusters to aid the coupling of multiple MGs. Meng et al. in Ref. [9] network. Each MG can also take advantage of the increased reliability
perform a review of control and energy-management schemes for MGs and stability, enhanced security and reduced costs associated with the
and clusters of multiple MGs, predominantly dealing with the control of cooperation among multiple MGs [8]. During a blackout of the utility
individual MG systems. The work in Ref. [10] provides a review of grid, individual MGs may fail to deliver a reliable and stable power
several challenges associated with the deployment of MGs and addresses supply to the critical loads due to the high penetration of
some of the literature contributions made regarding MG clusters. Alam non-controllable and weather-dependent RES. This can be overcome by
et al. in Ref. [11] present an extensive review of the works presented in interconnecting multiple MG systems and enabling them to support each
the literature regarding the operation of networked MGs. This research other by sharing resources during emergency situations [12,13]. For
presents a review of electrical protection, cooperation, and operational instance, MGs with abundant energy can supply their excess energy to
aspects of MG clusters, not comprehensively addressed in the existing the critical loads of MGs lacking in generation and stored energy.
technical literature. This includes control and energy-management However, the coupling of multiple systems requires the design and

2
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

development of control and energy-management architectures to allow 2.1. Non-cooperative game theory
safe and reliable operation of the entire MG cluster during
grid-connected and islanded mode, as well as new energy-trading Non-cooperative game theory is used in strategic decision-making
techniques to ensure all the potential benefits from the integration of processes involving different players. It is usually associated with so­
multiple MGs. Fig. 1 shows a brief diagram of an MG cluster containing cial phenomena in which there is some kind of conflict and it is not
four distinct interconnected MGs connected to the utility grid, in which possible to combine strategies or establish agreements between players.
each square of land shown inside the MG cluster represents an individual However, this does not always mean that players cannot cooperate with
MG. one another, it simply implies that there is no coordination or commu­
The development of clusters of MGs has been extensively discussed nication between players in the strategic decision-making processes.
and several energy-management frameworks have been proposed in the Non-cooperative game theory has been applied in smart grid and MG
literature [9]. In Ref. [14], the operation of MGs and MG clusters is applications in the context of energy markets, dynamic pricing, real-
analysed in both low-voltage and medium-voltage networks using time monitoring, and demand-side management [20]. Chen and Zhu
electricity market prices. This study shows that MG clusters lead to in Ref. [21] use a game-theoretic framework based on non-cooperative
economic benefits in terms of end-users cost reduction, active losses game to study the strategic behaviour of multiple MGs in a power dis­
elimination and emission reduction. Vasiljevska et al. in Ref. [15] tribution system while taking into consideration economic factors and
describe control and management functionalities to guarantee profit technical factors such as stability and efficiency. Liu et al. in Ref. [22]
from microgeneration units and loads in clusters of MGs. This is made propose a distributed coordination control scheme to address energy
through multi-criteria decision aid techniques to evaluate the impact of trading between multiple MGs in an electricity market environment
deploying clusters of MGs and select the best deployment strategy based using non-cooperative game theory. In this case, the use of a control
on a cost analysis. Liu et al. in Ref. [16] suggest a new real-time pricing scheme based on cooperative game would not be viable, because the
scheme based on the deployment of clustering techniques to allow each MGs have multiple coexisting beneficiaries and it may be very difficult
MG operator to assign real-time prices to the load patterns of each to reach an agreement among them.
customer in the cluster. This also helps in identifying the type of con­
sumer and their respective load curves in order to define the best de­ 2.2. Cooperative game theory
mand response programs for each consumer. In Ref. [17], an
agent-based intelligent energy-management system is proposed to While the outcome of non-cooperative games is defined by the ac­
facilitate the trade of energy among MGs and promote demand response tions of each individual player and by the payoff each player achieves,
programs. This study uses a system with two MGs and introduces the the outcome of cooperative games is defined by the forms of collabo­
concept of priority index for customers participating in demand ration between players within a group (i.e. coalitions), and its cooper­
response in accordance with the size and number of participation times. ative and self-organization actions. In cooperative game theory, players
The results showed a reduction in peak demand depending on the decide to form coalitions in order to make more effective decisions and
amount of load participation in demand response during peak times and maximize their benefits. Frameworks of cooperative game theory have
customers with higher priority index were able to obtain power supply been applied in the concept of smart grids to study and analyse the
at lower market prices. Nikmehr et al. in Ref. [18] study the benefit and cooperative behaviour among groups of MGs [23,24]. Prete and Hobbs
impact of demand response programs for optimal scheduling of a in Ref. [25] analyse the price, cost and benefit of the MG development
network of multiple MGs containing several types of RES under a new using a cooperative game theory framework. The authors conclude that
management system, while addressing the drawbacks of conventional utility customers can obtain most of the benefits associated with the
energy-management systems. Moreover [19], addresses two MGs oper­ development of MGs. In Ref. [26], a cluster of multiple independent MGs
ating in isolation from the utility grid and exchange energy between is considered and the price competition among these interconnected
each other using a peer-to-peer (P2P) architecture to minimize the total MGs is analysed using a game theory framework in which Nash Equi­
power generation and transportation costs. librium is computed and its uniqueness is shown. Du et al. in Ref. [27]
present a coalitional model for the cooperation and operation of mul­
tiple grid-connected MGs using the concept of cooperative game theory

Fig. 1. Physical layout of an MG cluster containing four distinct interconnected MGs connected to the utility grid: (1) Community MG; (2) Commercial MG; (3)
Residential MG; (4) Industrial MG.

3
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

to ensure a fair cost share among the MGs, and thus the economic sta­ advantages of implementing a system of systems energy management.
bility of the coalition. Korres et al. in Ref. [40] propose a state estimator based on weighted
least square (WLS) algorithm using few real-time measurements at the
3. Control and energy management transformer substation and distributed generation (DG) units, and
pseudo measurements for consumer load estimation. The MG cluster
The control and energy management of interconnected MGs can control scheme is required to operate with distinct control levels using
follow different architectures in order to ensure the optimized and co­ local microsource controllers (MC), load controllers (LC), an MG central
ordinated operation of each system during grid-connected and islanded controller (MGCC) and a central autonomous management controller
modes while guaranteeing minimal costs of operation. These architec­ (CAMC) serving as an interface to the distribution management system
tures can be centralized, decentralized, distributed or hierarchical. (DMS). Moreover, Madureira et al. in Ref. [41] propose a control
Other variations of the control and energy-management architecture strategy for the operation of MG clusters based on a hierarchical and
include the use of multi-agent technology or system-of-systems struc­ decentralized control scheme to ensure system autonomy and redun­
tures. Usually, the control of MG clusters is organized in multilayers or dancy. In addition to the coordinated voltage and frequency support
in a hierarchical hybrid fashion with certain control layers being functionality of the CAMC, the robustness of the proposed scheme pro­
implemented in a centralized or decentralized manner. vides an efficient use for local state estimation. Vasiljevska et al. in
Most of the control and energy-management schemes require two- Ref. [42] present a control functionality to manage DG units, energy
way communication links to transmit data between the various con­ storage devices and active loads in clusters of MGs to prevent overload
trollers, sensors and devices in the network. Reliable and resilient and voltage drop in the medium-voltage network. This functionality
communication can be achieved through a broad spectrum of wired and deals with inter-temporal relations of the energy-storage devices and
wireless communication technologies [28]. Wired technologies include active loads which are subjected to different constrains. In Ref. [43], a
power-line communications (PLC), digital subscriber lines (DSL) and multi-objective optimal power flow algorithm is presented to improve
optical fibre, while wireless technologies include Wi-Fi, WiMAX, the performance of MG clusters using interline power flow controller
Cellular 3G/4G, ZigBee and Bluetooth [29,30]. The choice of appro­ (IPFC). The proposed algorithm considers the control variables of the
priate technology is mostly dependent on characteristics such as IPFC and minimizes the operation costs of the MG cluster, energy losses
coverage area, data rate and operating frequency [28]. In smart grid and and bus voltage profile deviations. In Ref. [44], a coordinated optimal
MG applications, ensuring acceptable levels of quality-of-service (QoS) control algorithm modelled as a decentralized partially-observable
in terms of latency, capacity, reliability and availability is a difficult and Markov decision process (POMDP) is presented to coordinate the oper­
demanding task due to the complexity of data traffic in these networks. ation of networks containing multiple MGs. In Ref. [45], a coordinated
A wireless mesh network (WMN) topology with efficient routing pro­ control strategy is proposed for the coordination of multiple networked
tocols (i.e. multi-hop) and QoS mechanisms can be suitable for this type MGs and a DSO with distinct objectives using stochastic formulation.
of application and provide QoS guarantees to the network [30,31]. Recently, Ghomali et al. in Ref. [46] propose and demonstrate for the
During the last decade, several works have been presented in the first time in the literature the effectiveness of a comprehensive optimi­
literature to address the challenges of control and energy management zation and real-time control framework using an algorithm based on
in developing clusters of MGs [7]. The existing works often focus on mixed-integer second-order cone programming (MISOCP). Further­
economic aspects of the operation of MG clusters and on control stra­ more, several algorithms based on alternating direction method of
tegies which coordinate their operation. In Ref. [32], a multipliers (ADMM) have been proposed for the decentralized and
multi-agent-based centralized-decentralized hybrid hierarchical energy distributed energy management of networked MGs [47,48].
management is proposed for the optimization of distribution systems A comparison of typical control schemes for energy management in
containing multiple MGs. Kou et al. in Ref. [33] present a distributed clusters of MGs is presented in Table 1.
economic predictive control (EMPC) scheme for the coordinated sto­
chastic energy management of MG clusters, and compare it to the 3.1. Centralized
no-cooperation and centralized schemes. The proposed scheme showed
lower operating costs and a better balance between performance and In the centralized control approach, a single central controller de­
feasibility when compared to the no-cooperation and centralized termines the control actions of all local MCs and LCs in the system, as
schemes. In Ref. [34], a distributed optimal control framework is pro­ briefly represented in Fig. 2(a). The central controller coordinates the
posed in order to coordinate and achieve optimal control in clusters of optimal operation of the entire system with data gathered and received
MGs. This framework uses two control levels for each MG to optimize from the local units. Therefore, the operation of each MG system
the generation output of multiple MGs and to track the optimal gener­ participating in the cluster is controlled and coordinated by the central
ation setting of each DG unit during grid-connected and islanded oper­ controller. This control structure is simple and easy to implement due to
ation mode. In Ref. [35], a hierarchical control scheme is presented to the use of a single central controller [32]. However, large amounts of
solve the problem of controlling frequency and active power in a data are processed and confined in one location, resulting in a single
network containing multiple MG systems. In Ref. [36], a multi-agent point of failure which compromises the entire cluster operation [33].
control system for MG clusters is proposed to control and manage a Another downside is the fact that systems with centralized control
region containing multiple MGs. This control system allows the use of structures are often unable to expand into larger and more complex
plug-and-play and P2P capabilities to improve consumer participation networks. Moreover, because the central controller needs to receive data
and power system resiliency. In Ref. [37], a distributed control-oriented and send control signals to all local units, two-way communication is
hierarchical and distributed multi-agent system architecture is pre­ required between the central controller and each local unit [32,33].
sented as a reliable and flexible control scheme for clusters of MGs. In Song et al. in Ref. [49], propose an optimal energy-management method
Ref. [38], a system of systems framework is presented to optimally for a cooperative MG cluster integrated with a centralized control and
operate an active network containing a distribution company and mul­ energy-management system.
tiple MGs. The system of systems operation is modelled using a decen­
tralized optimal power flow problem and a hierarchical algorithm with 3.2. Decentralized
two control levels is applied to coordinate the operating points of all
systems in the network. Zhao et al. in Ref. [39] also propose a hierar­ In the decentralized control approach, each local controller de­
chical decentralized system of systems architecture for the optimized termines its own control actions and decisions based on data provided by
management of energy in a grid-connected MG cluster, and assess the local measurements, as briefly represented in Fig. 2(b). The local MCs

4
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Table 1 addition, security may be compromised since all data is shared among
Comparison of control schemes for energy management in clusters of MGs. multiple dispersed systems [33,34]. In Ref. [51], a distributed control
Advantages Disadvantages algorithm using ADMM is proposed for the energy management of a MG
cluster in order to minimize costs of operation. The proposed method
Centralized [32, • Offers a simple control • Unable to expand to larger
33] architecture with a single and more complex networks allows each MG to perform generation and load forecast independently
central controller • Requires communication without sharing important information regarding their operation.
• Low operation costs links between central
controller and local 3.4. Hierarchical
controllers
Decentralized • Suitable for the • Complicated control tasks
[32,33] coordination of multiple due to the lack of a central In a hierarchical control structure, the system is divided into multiple
autonomous systems controller control layers with distinct control tasks and goals, as represented in
• Provides autonomy to each • Decrease in aggregate Fig. 2(d). Therefore, the operation of each MG system participating in
individual system with no performance due to
the cluster is controlled and coordinated through multiple and distinct
communication among competitiveness among
systems individual systems control layers. For instance, the hierarchical control of a MG cluster may
Distributed [33, • Impacts to the aggregate • Data is shared among operate with multiple MCs and LCs in the first control layer, the MGCCs
34] performance are considered individual systems in the second control layer and the CAMC in the third control layer
and control tasks shared by compromising security which serves as an interface to the DMS [35]. This approach requires
each individual system • Requires communication
• Provides cooperation links between the
coordination methods between upper and lower layers, either in a more
among individual systems neighbouring systems centralized top-down hierarchy or in a less centralized bottom-up hi­
Hierarchical • Provides flexible layered • Susceptibility to failure due erarchy [32]. Because only adjacent layers can exchange data and units
[32,35] control of networks with to the strong dependency located at the same layer cannot share information with each other, all
multiple systems between lower and upper
the control decisions at the lower layer are strongly dependent on the
• Offers a cost-effective solu­ control levels
tion with easy implementa­ control state of the upper layer. This compromises the entire system in
tion and low operation costs the event of communication failure [32,35]. In Ref. [52], a two-layer
Multi-agent [36, • Solves the coordinated • Uses several dispersed hierarchical cooperative control method is proposed for networked
37] control problems by using agents that require MGs consisting of a distributed control layer for each MG to perform
autonomous and intelligent communication links to
agents interact with each other
frequency/voltage regulation and a decentralized control layer for the
• Offers network scalability entire MG cluster to enable the regulation of power flow through the
and plug-and-play PCC of each MG.
capabilities
System of • Aims to achieve the overall • High operation costs
3.5. Multi-agent
systems [38, goal of a multi-system
39] structure
• Offers flexibility and Distributed control based on multi-agent technology includes mul­
adaptability to a range of tiple autonomous and intelligent agents dispersed in the system. Unlike
decentralized controls the conventional agents of the distributed control scheme, these
advanced control agents can solve much more complex coordination
and LCs operate autonomously and coordinate their optimal operation problems in order to achieve a common or individual goal [37]. This
without relying on communication among each other. Therefore, each approach also offers high network scalability and enables plug-and-play
MG system participating in the cluster is responsible for gathering local capabilities. However, all the control agents in the scheme are inter­
data in order to coordinate its optimal operation. This approach offers linked, requiring communication links to interact with each other [36,
autonomy and increased stability to each unit while providing robust­ 37]. Zhai and Wang in Ref. [53] provide a multi-agent consensus algo­
ness against communication failures [32]. A decentralized control rithm to solve the real-time cooperative power dispatch problem of
structure also allows systems to easily scale up in size and complexity. islanded MG clusters and minimize costs associated with power
However, this approach cannot ensure the optimal operation of the regulation.
entire cluster, due to the absence of communication links and the overall
performance may be jeopardized by the increased competitiveness 3.6. System of systems
among individual units [32,33]. In Ref. [50], a decentralized
energy-management system is proposed for the coordinated operation of Control based on a system-of-systems structure integrates and co­
MG clusters during both grid-connected and islanded mode in a distri­ ordinates several autonomous systems to form a larger and more com­
bution system where the DSO and each MG are considered to constitute plex system capable of achieving a goal or task which could not be
distinct entities with individual goals. achieved by a single system or a group of uncoordinated systems [39].
All the systems are connected and coordinatively share their resources in
order to optimize the entire cluster operation and achieve the overall
3.3. Distributed goal. While this approach offers flexibility to a wide range of decen­
tralized controls, it is associated with high operational costs [38,39]. In
As depicted in Fig. 2(c), the distributed control approach relies on Ref. [54], a centralized model based on a system-of-systems structure is
various centralized controllers or central agents to coordinate the proposed to achieve optimal energy management of a cooperative MG
operation of the local units. Each central agent uses data provided by cluster.
other neighbouring agents and information from local MCs and LCs.
Hence, the operation of each in MG systems participating in the cluster is 4. Electrical protection schemes
controlled and coordinated by multiple dispersed central agents which
take account of the aggregate performance. This mitigates the compet­ Setting the appropriate electrical protections in MG systems to
itiveness among individual systems. In this approach, the coordination ensure safe and reliable operation in both grid-connected and islanded
among distributed agents allows the optimal performance of the entire modes can be problematic [55]. While traditional distribution networks
cluster to be achieved [33]. However, two-way communication is contain only unidirectional power flows, the MG protections must be
required to allow data to be shared between individual agents. In suited to deal with bidirectional power flow. This issue becomes much

5
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 2. Control schemes for MG clusters: (a) Centralized; (b) Decentralized; (c) Distributed; (d) Hierarchical.

more severe when operating a cluster of MGs, in which bidirectional protection systems [61]. The IEEE also recommends a practice for MG
power flow occurs between the utility grid and the cluster of MGs as well electrical system design, safety, power quality monitoring and control,
as among neighbouring MGs. The connection of several DG units electric energy measurement and scheme evaluation [62]. Even though
considered to be of intermittent nature, such as photovoltaic (PV) sys­ literature contributions for the electrical protection of MG clusters are
tems and wind turbines (WT), may cause fluctuating fault current levels scarce, a few studies can be found in the literature regarding the elec­
during islanded operation mode [56]. Moreover, MGs operating in trical protection of MG clusters. In Ref. [63], a fault analysis is per­
isolation from the utility grid produce relatively low fault currents, formed and reported for an MG cluster operating in islanded mode in
which may not be identified and detected by some protection devices. order to study the characteristics of different internal faults. The results
The use of protection devices without proper selectivity may also cause obtained from this study show that a cluster of MGs produces much
false tripping when the MG is operating in grid-connected mode [57,58]. higher fault currents than a single MG. In Ref. [64], a flexible protection
Therefore, conventional protection systems may not provide complete scheme is proposed for islanded MG clusters using controllers,
protection and cannot be implemented to solve the protection problems communication links and protection strategies such as differential pro­
of MGs and MG clusters during both grid-connected and islanded tection, directional protection and high impedance fault (HIF) detection.
operation mode. In the event of an external fault (e.g. fault in the utility This scheme isolated small faulty sections of the MG cluster and adapted
grid), the system needs to operate in islanded mode and provide a to the increased fault currents associated with the interconnection of
reliable supply of power to the critical loads. Generally, in the event of multiple MGs. Abdali et al. in Ref. [65] propose a protection scheme
an internal fault (e.g. fault in one of the MG feeders), the desired integrated with control interface based on fuzzy inference system (FIS)
response is to quickly isolate the smallest section possible in order to protection for MG interconnected systems. This method enables fast
enable the non-faulty sections to remain in operation [1]. fault detection, reduces protection costs and allows switching equip­
Numerous electrical protection schemes have been proposed in the ment with lower rated insulation capacity due to reduced switch peak
literature focusing on the protection of individual AC and DC MGs in voltage when compared to the differential protection. Yaqobi et al. in
both grid-connected and islanded operation mode [59,60]. An IEC Ref. [66] propose the integration of bidirectional solid-state circuit
standard on AC MGs specifies some technical requirements for DER in breakers (SSCB) based on insulated-gate bipolar transistors (IGBT) for
MGs, for MG connection to the distribution system, and for electrical high-speed protection of an isolated MG cluster. Recently, Zhang and Mu

6
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

in Ref. [67] suggest a protection scheme to identify and locate internal Table 3
faults in MG clusters based on voltage and current distribution charac­ Comparison of interconnection layouts, power conversion technologies and
teristics of the line parameters. The location of the fault is determined by power line frequency technologies for clusters of MGs.
comparing the phase differences and amplitudes of the measured feeder Advantages Disadvantages
admittances. Table 2 summarizes the electrical protection schemes Interconnection layout [69,70]
proposed in the literature for clusters of MGs. Parallel • Enables utility grid to be • Requires individual systems
supported by each individual to be self-sufficient during
5. Power system architecture system and offers increased faults in main distribution
reliability during faults grid
Series • Allows individual systems to • High susceptibility to
A network containing a power utility grid and multiple MGs can be support each other during stability issues when
arranged in different configurations [11,68]. Each MG in the region can faults in main distribution operating disconnected from
be connected to the utility grid through its PCC in a cluster of parallel grid main distribution grid
MGs, or all the MGs can be interconnected in a cluster of series inter­ Parallel • Allows individual systems to • Complex system with high
series support utility grid and/or protection and
connected MGs. Both configurations can be combined in a cluster of each other during emergency communication requirements
parallel-series MGs. In this case, some MGs are connected to the utility situations
grid, while others are interconnected in series followed by at least one Power conversion technology [69]
PCC to the utility grid [69,70]. In addition, the choice of interconnection Transformer • Provides reliable and cost- • Low controllability and
effective AC/AC voltage operation performance
and power line frequency technology needs to be taken into consider­
transformation
ation. MGs can be interconnected through power transformers in AC Converter • Offers good operation • High protection and
lines or through power converters in AC and DC lines. Power trans­ performance with active communication requirements
formers normally provide a reliable, cost-effective and non-controllable voltage regulation and high with high system costs
AC/AC voltage transformation in the interface of the utility grid and the controllability
Power line frequency technology [69,71,72]
MGs. On the other hand, most power converters offer an interconnection
AC line • Provides transformer • Requires P/Q control and
with a high level of controllability over AC or DC parameters [69]. In compatibility and uses mature frequency monitoring
Refs. [71,72], the performance of AC and DC technology in MG systems protection techniques
is addressed and discussed. Table 3 presents a brief comparison of DC line • Offers increased efficiency • High costs in terminal
and reduced losses due to equipment and difficulty in
interconnection layouts, power conversion technologies and power line
constant currents and fewer dealing with fault currents
frequency technologies for clusters of MGs [69–72]. conversion stages with low due to no zero-crossing point
Shahnia and Bourbour in Ref. [73] analyse various architectures of variety of faults and no
MG clusters and propose a suitable synchronization strategy to safely electromagnetic interference
interconnect neighbouring MGs. This strategy aims to decrease the
duration of abnormal voltage/frequency drops in overloaded MGs and
architecture for MG clusters which uses a hybrid unit of common
prevent the participating MGs from being temporarily overloaded dur­
coupling (HUCC) as an interface instead of the conventional PCC. This
ing the transition stage. Wu et al. in Ref. [74] present a novel
HUCC allows MGs to be connected to the utility grid via AC lines and
connected with each other via DC lines, which improves the flexibility
Table 2 and controllability of the MG cluster. In Ref. [75,76], back-to-back
Protection schemes proposed for clusters of MGs. (BTB) voltage source converters are used to interface interconnected
MGs. Furthermore, Goyal and Ghosh in Ref. [77] address the support
System Operation Study remarks Limitations
capabilities from the interconnection of two islanded MGs during con­
Gopalan [64] AC Islanded • Allows • Requires tingencies. This study shows the possibility of interconnecting an MG
et al. mode system to communication
isolate links to operate
operating in frequency-droop control with an MG operating in
small faulty angle-droop control through a BTB converter. Majumber and Bag in
sections and Ref. [78] suggest a control method to operate and coordinate parallel
adapt to MGs connected through BTB converters, allowing the autonomous
increased
operation of each MG without instabilities. Yoo et al. in Ref. [79] pro­
fault
currents pose the implementation of multi-frequency control in BTB converters in
Abdali [65] DC Grid- • Enables fast • Requires high- order to manage and improve the frequency regulation of isolated MG
et al. connected fault speed commu­ clusters. The proposed scheme for BTB converters offers a simple
and detection nication and approach for the regulation of multiple MG frequencies without
islanded and allows synchronization
mode equipment
communication requirements. Designs of flexible interconnection
with lower schemes and strategies for the interconnection of multiple MGs are
rated proposed and presented in Ref. [80,81].
insulation
capacity
6. Energy trading
Yaqobi [66] DC Islanded • Relatively • High cost
et al. mode fast short- associated with
circuit semiconductor Deploying several non-controllable RES with intermittent nature
interruption breakers results in unpredictable power generation. Certain customers may not be
technology able to generate enough power to supply their loads when the MG
Zhang [67] AC Grid- • Offers • High impedance
and connected simple and faults may not
cluster is operating in islanded mode. P2P energy trading allows cus­
Mu and flexible be detected in tomers to directly trade their excess energy with other customers and set
islanded fault load feeders their own terms and prices, without using a central entity as an inter­
mode detection mediary. This energy-trading scheme enables a fully decentralized and
with high
transparent internal energy market among the participating prosumers
sensitivity
and and customers. Both utility companies and grid costumers can take
reliability advantage of the benefits associated with energy trading in terms of

7
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

increased grid efficiency and reduced costs of grid operation [82]. In Cryptocurrency has also been proposed to allow a reliable and safe trade
Ref. [83], the feasibility of the cooperation between prosumers in of energy in smart grids [96].
establishing a participation in P2P energy trading within a network is
assessed, and a P2P energy-trading scheme is proposed using canonical 6.2. Energy-market designs
coalition game (CCG) to allow a coalition of prosumers to trade energy
with each other. In Ref. [84], the design, implementation and assess­ Recently, multiple projects on P2P energy trading have been devel­
ment of an energy market for energy trading in smart grid neighbour­ oped for prosumers and MGs in terms of energy-market designs, business
hoods is presented and discussed, showing the viability of implementing models, load control and ICT systems [97]. According to the literature,
this type of market. In Ref. [85], an energy-management optimization the existing prosumer energy markets can be categorized into fully
problem to minimize energy costs and P2P sharing losses in a network decentralized and less structured P2P models or into more structured
containing multiple MGs is proposed by transforming it into an equiv­ prosumer-to-grid (P2G) models [98,99]. A P2P energy market is a fully
alent second-order cone programming (SOCP) problem. This study decentralized model which allows all the participating peers (i.e. pro­
showed that the cooperation scheme in the P2P energy-trading network sumers and producers) to trade energy with each other directly,
is affected by the relative location of each MG cooperating. Zhang et al. providing an autonomous and flexible internal market. A P2G energy
in Ref. [86] identify and categorize important elements and technologies market is a structured model which relies on a community manager
associated with P2P energy trading, demonstrating potential benefits of controlling all the energy trades inside the community as well as any
this energy trading in MG systems. In Ref. [87], the energy trading interactions with other community managers in the grid (e.g. commu­
within an MG with P2P prosumers is studied. An energy-sharing model nities of prosumers connected to an MG) [98,99]. The P2P and the P2G
with priced-based demand response is proposed by formulating the in­ models can be combined in a hybrid market model. This hybrid model
ternal energy price model and cost model of prosumers. Recently, consists of prosumer groups which rely on a central entity such as a
Tushar et al. in Ref. [88] introduce a motivation psychology framework community manager controlling the energy trades, but these groups and
for smart grids consisting of several motivational models which each other peers can trade energy with each other in a P2P manner [98,100].
prosumer needs to satisfy in order to participate in energy trading. A P2P The typical structure and brief comparison of each prosumer
energy-trading scheme was also developed for this study using CCG, energy-market model according to Ref. [98], is presented in Fig. 3 and
demonstrating how a P2P energy-trading scheme can be designed and Table 4.
developed under a motivation psychology framework.
Several advanced information and communication technologies 7. Case study
(ICT) such as blockchain and cryptocurrency can provide the efficiency
and security required to benefit from P2P energy-trading markets in This section presents a case study to evaluate the cooperation among
smart grids and MG clusters. autonomous MGs participating in an MG cluster. This cooperation al­
lows each MG to share its resources with other neighbouring MGs in
6.1. Blockchain technology order to ensure a continuous supply of power to the critical loads during
islanded operation mode. Depending on the installed capacity and
Blockchain is an incorruptible digital ledger which records every location, certain MGs may be able to produce more than the amount
financial transaction permanently in a decentralized fashion [89]. they require to supply their critical loads. Taking this into account, these
Basically, it is composed of interconnected blocks where transactions are MGs with excess energy can supply the loads of neighbouring MG sys­
recorded. When completed, the block is stored in blockchain as a per­ tems with deficit energy if they are located in the same distribution grid
manent database and then a new block is generated. Each block in the [84,85]. The concept of MG clusters can play an important role in a
chain contains a reference to the previous block [90]. Blockchain was scenario where MGs with excess energy cooperate with MGs with deficit
originally created as the fabric for cryptocurrency, but its potential and energy by exchanging excess energy. To facilitate this energy trade
application in the energy market has significantly increased in the last among individual systems, an internal energy market needs to be
few years. In addition, blockchain technology allows digital information implemented to ensure transparency in the trade offers between sellers
to be duplicated thousands of times and shared among every peer, and buyers. This internal energy market can be either centralized or
meaning that data is stored and hosted in multiple points without a decentralized. Centralized market approaches differ from decentralized
single point of failure. Depending on the access to the ledger, blockchain approaches in such a way that a central entity known as the market
can be classified as either public (i.e. permission-less) or private (i.e. operator (MO) is necessary to establish market clearing prices and
permissioned) [90,91]. Public blockchains are completely decentralized quantities in each trading period (i.e. periods of 1 h) in accordance with
in nature and allow any anonymous entity to participate in the network, bids or offers made by the participating players. In decentralized market
while private blockchains only allow known and authorized entities to approaches such as in P2P energy trading, an MO is discarded and each
participate in the closed network governed by a single entity. Both types consumer trades directly with the suppliers [95]. Market clearing in
of blockchains rely on consensus mechanisms in order to verify and asymmetric pool requires bids from either demand or supply players and
validate the transactions. These consensus mechanisms include, among their respective amount of energy. In contrast, market clearing in sym­
others, proof-of-work (PoW), proof-of-stake (PoS) and metric pool requires bids from both demand and supply players. This
proof-of-authority (PoA) [90]. forms the social welfare area which is denoted by the area between the
The applications of blockchain technology in smart grids are demand and supply bid curves. The aggregation of the demand and
numerous. For instance, blockchain can facilitate automated billing supply bids follows a certain merit order. Demand bids are arranged in
systems, help identify consumer energy patterns, improve control of descending order from the highest bid to the lowest bid, while supply
decentralized systems and provide security and privacy to energy and bids are arranged in ascending order from the lowest bid to the highest
data transactions [92,93]. Mengelkamp et al. in Ref. [94] present the bid [101].
concept of an MG energy market based on blockchain technology The energy trade in the cluster during islanded operation mode is
without relying on a central entity to allow local energy trading in a P2P made using an internal market with hourly energy trading periods and
fashion among the participating agents. This energy market is success­ market clearing price/quantity in asymmetric pool. Therefore, three
fully applied to the real-world case of the Brooklyn MG. Wang et al. in market models are analysed in this case study, including an asymmetric
Ref. [95] propose a multidimensional willingness billing strategy to model for sell offers, an asymmetric model for buy offers, and a
allow P2P energy trading among MGs through a paralleled sequential asymmetric model for both sell and buy offers. An algorithm
energy-trading framework based on blockchain technology. is developed in Matlab, allowing coordination of the energy trade among

8
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 3. Prosumer energy markets: (a) P2P model; (b) P2G model; (c) Hybrid model.

installed renewable generation capacity, energy storage system (ESS)


Table 4
capacity and load profiles. The network is divided into two distinct
Comparison of prosumer market models (based on [98]).
zones. Zone 1 can operate both in radial or ring configuration by
Projects Advantages Disadvantages opening/closing switch S1 and consists of MG1, MG2 and MG3, while
Peer-to- • Piclo (United • Facilitates direct • Unable to guarantee zone 2 operates only in radial configuration and consists of MG4 and
peer Kingdom) financial safe and high-quality MG5. In this study, the MG cluster is disconnected from the main dis­
model • Vandebron transactions delivery of energy
tribution grid and operates autonomously for an entire day (i.e. 24 h).
(Netherlands) among individual • High
prosumers in fully implementation and Each MG is connected to the cluster through a MV/LV step-down
decentralized maintenance costs transformer. Fig. 4 represents the simplified single-line diagram of the
fashion due to diverse and 11-bus network containing the five MGs under study. The simulations
• Offers flexibility decentralized presented in the following sections are made with switch S1 in the closed
and autonomy to networks
state.
each prosumer and
customer The energy consumption under normal operation corresponds to the
Prosumer- • NOBEL • Promotes the • Difficulty integrating operation of each MG at full rated power. During islanded operation
to-grid (Europe) integration of and managing large mode, each MG reduces its energy consumption in order to ensure a
model multiple amounts of data from
continuous supply of power to the critical loads. Fig. 5 and Fig. 6 show
prosumers into numerous prosumers
cooperative
the load profile of each MG in normal and islanded operation sce­
communities narios for every hour of the day.
• Provides various
services and high-
quality energy de­ 7.2. Algorithm for energy trading in MG clusters
livery due to part­
nerships among
An algorithm allows multiple MGs to perform an exchange of excess
numerous individ­
ual prosumers
energy in order to compensate for the periods of time when one or more
Hybrid • REV (United • Aims to serve the • Difficulty integrating MGs require a supply of energy, allowing autonomous operation of the
model States) interests of and managing large entire cluster when disconnected from the main distribution grid (i.e.
communities, amounts of data from utility grid). The algorithm, developed in Matlab, determines the energy
organizations or prosumer groups
balance of each MG for every hour of the day. Fig. 7 presents a brief
groups of • Complex transaction
prosumers and management flowchart of the algorithm developed.
• Provides various costs within each The energy balance of each MG being studied is determined using the
services and high- group of prosumers input data available for each prosumer. Input data includes hourly PV
quality energy de­ generation, WT generation, ESS capacity and energy consumption. In
livery due to part­
nerships among
addition, input data also includes the price in €/kWh for which each MG
limited numbers of sells or buys excess energy. The next steps of the Matlab script use the
prosumer groups function “linprog” and the open-source package “MATPOWER” for the
three market clearing models. The function “linprog” allows the optimal
solution to be identified for a specific linear problem defined by
the MGs.
inequality constraints, equality constraints and boundaries. This func­
tion is structured in accordance with each market clearing model. The
7.1. MG cluster under study open-source package “MATPOWER” is used in the algorithm to allow the
analysis of power flow and determine losses in the network under study.
The MG cluster under study includes five industrial MGs dispersed in For this case study, the default case file has been used as template and
the network. Each MG corresponds to a distinct prosumer with different adapted in accordance with the parameters of the network under study.

9
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 4. Simplified single-line diagram of the MG cluster being studied.

Fig. 5. Load profile of each MG in normal operation.

In the algorithm developed, the matrices corresponding to each field are to supply their excess energy in accordance with their respective energy
updated in each time instant. This includes values of active and reactive market prices. The amount of power subjected to load-shedding is
power in loads and generators determined by the algorithm or imported associated with a high cost as a penalty for any non-compliance of the
from the database for each bus of the network. energy supply. The objective function for this problem is given by (1):
Model 1 – Sell offers. This first model consists of sell offers in n ( )

asymmetric pool made by MGs with excess energy. Therefore, each MG min ΔPGipv + ΔPGiwt + ΔPGib × CGi + PLshed × Cshed (1)
sets a specific price for its excess energy, and offers are ranked based on
i=1

ascending order of energy price as represented in Fig. 8. where ΔPGipv , ΔPGiwt and ΔPGib are the amount of excess power generated
In this model, MGs only ensure the supply of the critical loads,
by PV panels, WTs and stored in battery systems for each MG, CGi is their
meeting their minimal energy consumption. If the total excess energy
respective energy cost, and PLshed is the amount of power subjected to
cannot meet the total energy demand, load-shedding is applied to the
load-shedding which is associated with a high cost value Cshed . This high
MGs with deficit energy in a proportional manner. The main objective is
cost is a penalty for any non-compliance of the energy supply, being
to optimize the cost of the energy trade between individual MGs.
required to safeguard each market offer made by the MGs with excess
Therefore, if at least one MG requires a supply of energy from other
generation. This function is subjected to the following inequality
neighbouring MGs, the function “linprog” will select the best candidates
constraint (2):

10
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 6. Load profile of each MG in islanded operation.

ΔPGipv + ΔPGiwt + ΔPGib ≤ ΔPGitotal , i = 1, …, n (2) In other words, one or more MGs are selected to receive the excess en­
ergy at the highest possible cost. The objective function for this problem
where ΔPGitotal is the total amount of excess power generated available in is given by (5):
each MG, which can be determined by adding together all of the excess

n
power generated by PV panels, WTs and stored in battery systems. max ΔPLi × CLi − PGspill × Cspill (5)
The equality constraint of the problem is given by (3), as follows:
i=1


n
( ) where ΔPLi is the amount of deficit power needed to supply critical loads
ΔPGipv + ΔPGiwt + ΔPGib + PLshed = ΔPLtotal (3) in each MG, CLi is the respective energy cost, and PGspill is the amount of
i=1
power subjected to energy spill which is also associated with a high cost
where ΔPLtotal is the total amount of power consumed by all MGs value Cspill as a penalty. This function is subjected to the following
participating in the cluster. inequality constraints (6):
Finally, the objective function is also limited to both upper and down ΔPLi ≤ ΔPLitotal , i = 1, …, n (6)
boundaries given by (4):

ΔPGipv
⎤ ⎡
PGipv
⎤ where ΔPLitotal is the total amount of power needed to supply all the

0 ≤ ΔPGiwt ≤ PGiwt ⎦
⎦ ⎣ critical loads in each MG.
, i = 1, …, n (4)
ΔPGib PGib The equality constraint of the problem is given by (7), as follows:
0 ≤ PLshed ≤ ΔPLtotal

n
ΔPLi − PGspill = ΔPGtotal (7)
where PGipv , PGiwt and PGib are the amount of power generated by PV i=1
panels, WTs and stored in battery systems for each MG. These values are
directly imported from the database of each MG. The amount of load- where ΔPGtotal is the total amount of excess power generated available in
shedding cannot exceed the total amount of power consumed by the the MG cluster, which can be determined by adding together all of the
entire cluster and needs to be greater than or equal to zero. excess power generated by PV panels, WTs and stored in battery
The solution obtained from “linprog” corresponds to the amount of systems.
excess energy supplied by one or more MGs with the lowest energy cost Finally, the objective function is also limited to both upper and down
if there is at least one MG with deficit energy. These values are also boundaries given by (8):
required to update the state of the ESS in each MG for the next time 0 ≤ ΔPLi ≤ PLi
instant. Moreover, the solution also includes the amount of load needing , i = 1, …, n (8)
0 ≤ PGspill ≤ ΔPGtotal
to be shed when the total amount of deficit energy is larger than the total
amount of excess energy in the cluster. Then, market clearing price and where PLi is the amount of power required in each MG. These values are
quantity graphs in asymmetric pool for sell offers are obtained using the directly imported from the database of each MG. The amount of energy
amount of load-shedding, excess energy and deficit energy. spill cannot exceed the total amount of excess energy generated by the
Model 2 – Buy offers. This model consists of buy offers in asymmetric entire cluster and needs to be greater than or equal to zero.
pool made by MGs with deficit energy. Therefore, MGs with deficit en­ The solution obtained from “linprog” corresponds to the amount of
ergy set a price for the quantity of energy they require, and each offer is excess energy received by one or more MGs if there is at least one MG
ranked based on descending order of energy price, as represented in with excess energy. These values are also required to update the state of
Fig. 9. the ESS in each MG for the next time instant. Moreover, the solution also
In this model, MGs also must ensure the supply of the critical loads, includes the amount of energy spill needing to be spilled when the total
meeting their minimal energy consumption. If the total excess energy amount of excess energy is larger than the total amount of deficit energy
cannot meet the total energy demand, the excess energy is exchanged in the cluster. Then, market clearing price and quantity graphs in
with the MGs which offer the highest price for that amount of energy. asymmetric pool for buy offers are obtained using the amount of energy
The function “linprog” will select the best candidates to receive the spill, excess energy and deficit energy.
excess energy in accordance with their respective energy market prices. Model 3 – Sell or buy offers. This sequential model is a hybrid of the

11
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 7. Algorithm flowchart for model 1 and model 2.

Fig. 8. Sell offers ranked based on ascending order of energy price. Fig. 9. Buy offers ranked based on descending order of energy price.

12
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

two previous models. It consists of either sell or buy offers in asymmetric instants.
pool made by MGs with excess or deficit energy. If the total excess en­ Model 2 – Buy offers. The market clearing price and quantity graphs
ergy exceeds the total deficit energy, each MG sets a price for the unused for buy offers in the same time instants h = 15 and h = 22 are shown in
excess energy in order to increase its energy consumption beyond the Fig. 12. The power-flow results for this model in these time instants are
supply of critical loads and to attempt to meet its normal energy de­ also presented in Table 5.
mand. Like the previous model, buy offers are ranked based on As observed in the market clearing price and quantity graph for buy
descending order of energy price. The excess energy is exchanged with offers in time instant h = 15 shown in Fig. 12, two buy offers are rep­
the MGs which offer the highest price, as represented in Fig. 10. This resented for the two MGs with deficit energy. The vertical line in blue
ensures that any unused excess energy is not wasted, depending on the represents the total excess energy and corresponds to the sum of excess
operation cycle of each MG and whether batteries can be charged or not. energy for the three remaining MGs with excess energy in this time
On the other hand, if the total excess energy does not exceed the total instant. In this case, the total excess energy equals the amount of deficit
deficit energy, sell offers and energy exchanges are made pursuant to energy because the total excess energy is higher than the total deficit
model 1. energy. In the market clearing price and quantity graph for buy offers in
In this case, the algorithm follows the same function structure as time instant h = 22 also shown in Fig. 12, four buy offers are represented
model 1, which selects the best candidates to supply their excess energy for the four MGs with deficit energy. The vertical line in blue represents
based on their respective energy market prices to meet the total deficit the total excess energy and corresponds to the remaining MG with excess
energy in the MG cluster at the lowest possible cost. Then, the algorithm energy in this time instant. In this case, only one MG receives the amount
follows the function structure of model 2 if the amount of excess energy of excess energy available because the total excess energy cannot meet
exceeds the amount of deficit energy in the cluster. Therefore, one or the total deficit energy. For this model, the energy exchange price is 0.20
more MGs receive the extra excess energy in accordance with the €/kWh in time instant h = 15 and 0.30 €/kWh in time instant h = 22,
amount of energy required to meet its normal energy demand. both prices are higher than the prices in model 1 because model 2 is
based on buy offers. As shown in Table 5, the bus voltage is within ideal
7.3. Results values (approximately 1 pu) for all buses of the network. The trans­
mission losses are negligible in these time instants.
The results obtained for each asymmetric model are presented and Model 3 – Sell or buy offers. The market clearing price and quantity
discussed in the following sections for two distinct time instants. In this graphs for sell or buy offers in the same time instants h = 15 and h = 22
study, it is assumed that each industrial MG integrates a compensation of are shown in Fig. 13. The power-flow results for this model in these time
reactive energy to avoid payment of the respective invoice. Therefore, instants are also presented in Table 5.
the reactive energy consumption of each MG corresponds to a power As observed in the market clearing price and quantity graph shown
factor of 0.95 (cosφ ≥ 0.95 or tanφ ≤ 0.3). in Fig. 13, buy offers are represented for all the MGs in the time instant h
Model 1 – Sell offers. The market clearing price and quantity graphs = 15. This is because all MGs already ensured the supply of their critical
for sell offers in time instants h = 15 and h = 22 are shown in Fig. 11. loads, due to the total excess energy exceeding the total amount of
The power-flow results for this model in time instants h = 15 and h = 22 deficit energy. The buy offers represented in the graph are set in
are presented in Table 5. This includes magnitude and phase angle accordance with the amount of energy each MG needs in order to meet
values for the bus voltage. its normal energy demand. The vertical line in blue represents the total
As observed in the market clearing price and quantity graph for sell excess energy, and corresponds to the sum of excess energy of all the
offers in time instant h = 15 shown in Fig. 11, three sell offers are MGs with excess energy in this time instant. It is worth mentioning that
represented for the three MGs with excess energy. The vertical line in the time instant h = 22 presents the same results as model 1 because
blue represents the total deficit energy. It corresponds to the sum of model 3 follows the same structure as model 1 when the total excess
deficit energy for the two remaining MGs which have deficit energy in energy cannot meet the total energy demand. For both time instants, the
this time instant. In the market clearing price and quantity graph for sell energy exchange price corresponds to 0.15 €/kWh. Table 5 shows a bus
offers in time instant h = 22 also shown in Fig. 11, a sell offer is rep­ voltage within relatively ideal values (approximately 1 pu), with the
resented for the only MG with excess energy. In this case, load-shedding lowest voltage being 0.997 pu in bus 10 for time instant h = 15. There
is applied to any MG with deficit energy because the total excess energy are some transmission losses (around 1.64 kW) in time instant h = 15
cannot meet the total deficit energy. The vertical line in red represents due to an increased amount of energy being exchanged.
the total deficit energy posterior to load-shedding, corresponding to the
total excess energy. Therefore, MGs with deficit energy are subjected to 7.4. Discussion
a load decrease proportional to their amount of deficit energy, which is
represented as vertical dashes. For this model, the energy exchange price When comparing the three models previously presented, it is
corresponds to 0.15 €/kWh in both time instants. As shown in Table 5, noticeable that model 3 constitutes a relatively better and more com­
the bus voltage is within ideal values (approximately 1 pu) for all buses plete model than the other two models. The main advantage of adopting
of the network. The transmission losses are negligible in these time this model is the possibility of any unused excess energy to be exchanged

Fig. 10. Sell and buy offers ranked based on ascending and descending order of energy price.

13
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 11. Market clearing price and quantity graph for model 1 in time instants h = 15 and h = 22.

Table 5
Magnitude and phase angle values for the bus voltage in pu.
Model 1 Model 2 Model 3

Voltage (h = 15) Voltage (h = 22) Voltage (h = 15) Voltage (h = 22) Voltage (h = 15) Voltage (h = 22)

Bus Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang

1 1.001 0.213 1.001 0.017 1.001 0.213 1.001 0.000 1.002 2.108 1.001 0.017
2 1.001 0.213 1.001 0.017 1.001 0.213 1.001 − 0.000 1.002 2.107 1.001 0.017
3 1.000 0.000 1.000 0.000 1.000 0.000 1.000 0.000 1.000 0.000 1.000 0.000
4 1.001 0.214 1.001 0.017 1.001 0.214 1.001 − 0.000 1.002 2.118 1.001 0.017
5 1.000 0.214 1.000 0.014 1.000 0.214 1.000 − 0.049 1.000 5.629 1.000 0.014
6 1.001 0.215 1.001 0.017 1.001 0.215 1.001 0.000 1.002 2.117 1.001 0.017
7 1.000 0.854 1.000 0.066 1.000 0.854 1.000 0.049 1.000 3.170 1.000 0.066
8 1.001 0.199 1.001 0.022 1.001 0.199 1.001 0.006 0.999 1.999 1.001 0.022
9 1.000 0.199 1.000 0.014 1.000 0.199 1.000 0.006 1.000 1.161 1.000 0.014
10 1.000 0.183 1.001 0.024 1.000 0.183 1.001 0.009 0.997 1.297 1.001 0.024
11 1.000 − 0.244 1.000 0.003 1.000 − 0.244 1.000 0.009 1.000 0.310 1.000 0.003

Fig. 12. Market clearing price and quantity graph for model 2 in time instants h = 15 and h = 22.

and sold to the MGs with the highest buy price. This is dependent on the model 1 takes the cooperation among multiple MGs into consideration,
operation cycle of each MG and whether batteries can be charged or not. because it uses sell offers for the excess energy and this energy is
However, any extra excess energy can also be distributed to nearby distributed proportional to the MGs with deficit energy. On the other
buildings located in the same network if none of the MGs participating in hand, model 2 only supports the interests of individual MGs, because it
the cluster can receive any excess energy (i.e. batteries at full capacity or uses buy offers based on the energy needs of each MG, and only the MGs
MGs operating at full rated power). In the time instants with load- with the highest buy price receive the excess energy. However, an
shedding, model 1 applies the excess energy to the MGs with deficit approach similar to model 1 which distributes excess energy in a pro­
energy in a proportional manner, while model 2 applies the excess en­ portional manner to the MGs with deficit energy could be applied to
ergy to the MGs with deficit energy and highest buy price. In such a way, model 2 to support the cooperation among multiple MGs. In this

14
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Fig. 13. Market clearing price and quantity graph for model 3 in time instants h = 15 and h = 22.

approach, MGs would need to set a pre-defined buy price for when load- of energy supplied [102].
shedding is applied. For instance, this buy price can be equal to the
utility energy price in order to be as transparent as possible to each MG. 8. Conclusions
Another approach for model 2 or 3 could consist of setting pre-defined
buy prices for the excess energy to be supplied to loads with different This paper addresses the control, architecture and energy trade in
priority, as represented in Fig. 14. These buy offers for specific types of clusters of MGs. To this end, this work provided a review of the most
loads would be ranked based on descending order of energy price. For relevant research made towards the project and the development of
instance, MGs can set a high buy price for the amount of energy smart grids. The research presented in the literature mainly focuses on
necessary to supply all the loads with high priority (i.e. critical and the control and energy management of MG clusters, with various control
essential loads), a medium buy price for energy necessary to supply strategies and algorithms being proposed to guarantee the optimal
loads with medium priority (i.e. non-critical loads but essential to pre­ operation and economic feasibility of the system. These control strate­
vent financial loss), and a low buy price for any energy necessary to gies can follow centralized, decentralized and distributed architectures
supply the remaining loads. or be implemented in layers with different tasks and goals through a
Payment for power transactions in the cluster of MGs can be made hierarchical architecture. Multi-agent technology and system-of-systems
using virtual currency (i.e. cryptocurrency) or conventional payments. structures have also been proposed in the literature for the control and
Using cryptocurrency makes it possible to have almost instantaneous energy management of multiple MG systems. A comparison of these
payments through virtual transactions. These payments may be made by control and energy-management architectures has been provided in this
the use of a public blockchain (e.g. Ethereum or Bitcoin). In such case, work. The architecture of MG clusters in terms of layout, power con­
the commercialization of P2P power will be implemented as a sidechain version technology and line frequency technology has also been
(i.e. validation of data between blockchains) [96,102]. Despite its easy addressed. Moreover, P2P energy markets are another focus of the
implementation, the floating exchange rate for the cryptocurrency may research presented in the literature. This work has briefly dealt with
be an obstacle. The conventional payment solution consists of remu­ three distinct energy market designs which are suitable for imple­
neration of MGs by utilities. In this case, utilities can invest in the con­ mentation in MG clusters to facilitate and optimize the energy trade
struction of P2P power grids, acting as operators of the power system. among the participating prosumers, producers and customers. Finally, in
The MGs participating in the P2P energy trade are still able to maintain the field of protections, numerous electrical protection schemes have
their interconnection to the traditional power distribution grid. In this been proposed in the literature concentrating on the protection of in­
type of configuration, the MGs are affiliated with the utility company. dividual MGs in both grid-connected and islanded operation mode.
This makes it possible for an MG to supply power to another MG on However, literature contributions for the electrical protection of MG
behalf of the utility, and payment to the MGs is based on the net quantity clusters are still scarce. It is worth mentioning that the scope of this
paper does not cover cybersecurity aspects such as privacy issues and
data protection against cyber-attacks in MG clusters. Nonetheless, this is
a very interesting and important topic, and may be considered for future
research work.
The study carried out in this paper has addressed the cooperation of
five industrial MGs operating in islanded mode for an entire day. An
algorithm has been developed in Matlab to evaluate the cooperation
among these MGs and coordinate their hourly energy trade through
three distinct market models with market clearing price/quantity in
asymmetric pool. The results show that a proportional distribution of
excess energy to the MGs with deficit energy is the most suitable in a
cooperative environment, while the supply of excess energy to the MGs
with the highest buy price only supports the interests of these individual
MGs.

Fig. 14. Excess energy prices in accordance with loads with different priority.

15
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

Declaration of competing interest [24] Camarinha-Matos LM. Collaborative smart grids–A survey on trends. Renew
Sustain Energy Rev 2016;65:283–94.
[25] Prete CL, Hobbs BF. A cooperative game theoretic analysis of incentives for
The authors declare that they have no known competing financial microgrids in regulated electricity markets. Appl Energy 2016;169:524–41.
interests or personal relationships that could have appeared to influence [26] Kasbekar GS, Sarkar S. Pricing games among interconnected microgrids. In:
the work reported in this paper. Proceedings of the IEEE PES general meeting, San Diego, California, USA; 2012.
p. 1–8.
[27] Du Y, Wang Z, Liu G, Chen X, Yuan H, Wei Y, Li F. A cooperative game approach
Acknowledgements for coordinating multi-microgrid operation within distribution systems. Appl
Energy 2018;222:383–95.
[28] Bandeiras F, Gomes M, Coelho P, Fernandes J, Camacho A, Castilla M. Microgrid
This work was partially supported by the Portuguese Foundation for architecture evaluation for small and medium size industries. Int J Emerg Elec
Science and Technology (FCT) and by PIDDAC, under the research Power Syst 2018;19(2).
project INDuGRID, ERANETLAC/0006/2014. [29] Tsado Y, Gamage K, Lund D. Communication technologies for smart grid
ubiquitous sensor network system. In: Proceedings of the 4th international
Trina Cairns (native English speaker and professional proofreader), conference on power engineering, energy and electrical drives. Turkey: Istanbul;
who reviewed the text of this work. 2013. p. 1706–12.
[30] Tsado Y, Lund D, Gamage K. Resilient wireless communication networking for
smart grid BAN. In: Proceedings of the 2014 IEEE international energy conference
References (ENERGYCON), cavtat, Croatia; 2014. p. 846–51.
[31] Tsado Y, Gamage KAA, Adebisi B, Lund D, Rabie KM, Ikpehai A. Improving the
[1] Bandeiras F, Gomes M, Coelho P, Fernandes J, Moreira C. Protection systems and reliability of optimised link state routing in a smart grid neighbour area network
earthing schemes for microgrids: main aspects and fault analysis. Int J Emerg Elec based wireless mesh network using multiple metrics. Energies 2017;10(3).
Power Syst 2018;19(4). [32] Mao M, Wang Y, Chang L, Du Y. Operation optimization for multi-microgrids
[2] Moslehi K, Kumar R. A reliability perspective of the smart grid. IEEE Trans Smart based on centralized-decentralized hybrid hierarchical energy management. In:
Grid 2010;1(1):57–64. Proceedings of the 2017 IEEE energy conversion congress and exposition.
[3] Hirsch A, Parag Y, Guerrero J. Microgrids: a review of technologies, key drivers, Cincinnati, Ohio, USA: ECCE); 2017. p. 4813–20.
and outstanding issues. Renew Sustain Energy Rev 2018;90:402–11. [33] Kou P, Liang D, Gao L. Distributed EMPC of multiple microgrids for coordinated
[4] Microgrids for commercial and industrial companies: delivering increased power stochastic energy management. Appl Energy 2017;185:939–52.
reliability, lower energy costs and lower emissions. World Business Council for [34] Zhang W, Xu Y. Distributed optimal control for multiple microgrids in a
Sustainable Development (WBCSD); 2017. Available online: https://docs.wbcsd. distribution network. IEEE Trans Smart Grid 2019;10(4):3765–79.
org/2017/11/WBCSD_microgrid_INTERACTIVE.pdf. [Accessed 11 May 2018]. [35] Gil NJ, Lopes JAP. Hierarchical frequency control scheme for islanded multi-
[5] Insights Guidehouse. Microgrid deployment tracker 1Q20. Available online: htt microgrids operation. In: Proceedings of the 2007 IEEE lausanne power tech,
ps://guidehouseinsights.com/reports/microgrid-deployment-tracker-1q20. lausanne, Switzerland; 2007. p. 1–6.
[Accessed 20 July 2020]. [36] Ng EJ, El-Shatshat RA. Multi-microgrid control systems (MMCS). In: Proceedings
[6] Schmitt L, Kumar J, Sun D, Kayal S, Venkata SS. Ecocity upon a hill: microgrids of the IEEE PES general meeting, providence. USA: Rhode Island; 2010. p. 1–6.
and the future of the European city. IEEE Power Energy Mag 2013;11(4):59–70. [37] Zheng G, Li N. Multi-agent based control system for multi-microgrids. In:
[7] Zafar R, Mahmood A, Razzaq S, Ali W, Naeem U, Shehzad K. Prosumer based Proceedings of the 2010 international conference on computational intelligence
energy management and sharing in smart grid. Renew Sustain Energy Rev 2018; and software engineering. Wuhan: RPC; 2010. p. 1–4.
82:1675–84. [38] Marvasti AK, Fu Y, DorMohammadi S, Rais-Rohani M. Optimal operation of
[8] Espe E, Potdar V, Chang E. Prosumer communities and relationships in smart active distribution grids: a system of systems framework. IEEE Trans Smart Grid
grids: a literature review, evolution and future directions. Energies 2018;11(10). 2014;5(3):1228–37.
[9] Meng L, Shafiee Q, Trecate GF, Karimi H, Fulwani D, Lu X, Guerrero JM. Review [39] Zhao B, Wang X, Lin D, Calvin MM, Morgan JC, Qin R, Wang C. Energy
on control of DC microgrids and multiple microgrid clusters. IEEE Trans Emerg management of multiple microgrids based on a system of systems architecture.
Sel Topics Power Electron 2017;5(3):928–48. IEEE Trans Smart Grid 2018;33(6):6410–21.
[10] Parhizi S, Lotfi H, Khodaei A, Bahramirad S. State of the art in research on [40] Korres GN, Hatziargyriou ND, Katsikas PJ. State estimation in multi-microgrids.
microgrids: a review. IEEE Access 2015;3. Eur Trans Electr Power 2011;21(2):1178–99.
[11] Alam MN, Chakrabarti S, Ghosh A. Networked microgrids: state-of-the-art and [41] Madureira AG, Pereira JC, Gil NJ, Lopes JAP, Korres GN, Hatziargyriou ND.
future perspectives. IEEE Trans. Ind. Inf. 2019;15(3):1238–50. Advanced control and management functionalities for multi-microgrids. Eur
[12] Saleh MS, Althaibani A, Esa Y, Mhandi Y, Mohamed AA. Impact of clustering Trans Electr Power 2011;21(2):1159–77.
microgrids on their stability and resilience during blackouts. In: Proceedings of [42] Vasiljevska J, Lopes JAP, Matos MA. Integrated micro-generation, load and
the 2015 international conference on smart grid and clean energy technologies energy storage control functionality under the multi micro-grid concept. Elec
(ICSGCE), Offenburg, Germany; 2015. p. 195–200. Power Syst Res 2013;95:292–301.
[13] Arnett A, Putman E, Dempsey M. Enabling a grid of microgrids. Analysis and [43] Kargarian A, Falahati B, Fu Y, Baradar M. Multiobjective optimal power flow
policy observatory (APO). 2018. Available online: https://apo.org.au/sites/defau algorithm to enhance multi-microgrids performance incorporating IPFC. In:
lt/files/resource-files/2018/05/apo-nid175201-1104751.pdf. [Accessed 10 Proceedings of the IEEE PES general meeting, San Diego, California, USA; 2012.
January 2019]. p. 1–6.
[14] Anastasiadis AG, Tsikalakis AG, Hatziargyriou ND. Operational and [44] Wu J, Guan X. Coordinated multi-microgrids optimal control algorithm for smart
environmental benefits due to significant penetration of microgrids and topology distribution management system. IEEE Trans Smart Grid 2013;4(4):2174–81.
sensitivity. In: Proceedings of the IEEE PES general meeting, providence. USA: [45] Wang Z, Chen B, Wang J, Begovic MM, Chen C. Coordinated energy management
Rhode Island; 2010. p. 1–8. of networked microgrids in distribution systems. IEEE Trans Smart Grid 2015;6
[15] Vasiljevska J, Lopes JAP, Matos MA. Evaluating the impacts of the multi- (1):45–53.
microgrid concept using multicriteria decision aid. Elec Power Syst Res 2012;91: [46] Gholami A, Sun XA. Towards resilient operation of multi-microgrids: an MISOCP-
44–51. based frequency-constrained approach. IEEE Trans Control Netw Syst 2019;6(3):
[16] Liu H, Mahmoudi N, Chen K. Microgrids real-time pricing based on clustering 925–36.
techniques. Energies 2018;11(6). [47] Gao H, Liu J, Wang L, Wei Z. Decentralized energy management for networked
[17] Nunna H, Doolla S. Demand response in smart distribution system with multiple microgrids in future distribution systems. IEEE Trans Power Syst 2018;33(4):
microgrids. IEEE Trans Smart Grid 2012;3(4):1641–9. 3599–610.
[18] Nikmehr N, Najafi-Ravadanegh S, Khodaei A. Probabilistic optimal scheduling of [48] Ma W-J, Wang J, Gupta V, Chen C. Distributed energy management for
networked microgrids considering time-based demand response programs under networked microgrids using online ADMM with regret. IEEE Trans Smart Grid
uncertainty. Appl Energy 2017;198:267–79. 2018;9(2):847–56.
[19] Matamoros J, Gregoratti D, Dohler M. Microgrids energy trading in islanding [49] Song N-O, Lee J-H, Kim H-M, Im YH, Lee JY. Optimal energy management of
mode. In: Proceedings of the IEEE SmartGridComm 2012 symposium. Tainan multi-microgrids with sequentially coordinated operations. Energies 2015;8(8):
Taiwan; 2012. p. 49–54. 8371–90.
[20] Saad W, Han Z, Poor H, Basar T. Game-theoretic methods for the smart grid: an [50] Wang Z, Chen B, Wang J, Kim J. Decentralized energy management system for
overview of microgrid systems, demand-side management, and smart grid networked microgrids in grid-connected and islanded modes. IEEE Trans Smart
communications. IEEE Signal Process Mag 2012;29(5):86–105. Grid 2016;7(2):1097–105.
[21] Chen J, Zhu Q. A game-theoretic framework for resilient and distributed [51] Liu Y, Gooi HB, Xin H. Distributed energy management for the multi-microgrid
generation control of renewable energies in microgrids. IEEE Trans Smart Grid system based on ADMM. In: Proceedings of the IEEE PES general meeting,
2017;8(1):285–95. chicago, Illinois, USA; 2017. p. 1–5.
[22] Liu W, Gu W, Wang J, Yu W, Xi X. Game theoretic non-cooperative distributed [52] Wu X, Xu Y, Wu X, He J, Guerrero JM, Liu C-C, et al. A two-layer distributed
coordination control for multi-microgrids. IEEE Trans Smart Grid 2018;9(6): cooperative control method for islanded networked microgrid systems. IEEE
6986–97. Trans Smart Grid 2020;11(2):942–57. https://doi.org/10.1109/
[23] Saad W, Han Z, Poor H. Coalitional game theory for cooperative micro-grid TSG.2019.2928330.
distribution networks. In: Proceedings of the IEEE international conference on [53] Zhai X, Wang N. Multi-agent consensus algorithm-based optimal power dispatch
communications (ICC), Kyoto, Japan; 2011. p. 1–5. for islanded multi-microgrids. Processes 2019;7(10).

16
F. Bandeiras et al. Renewable and Sustainable Energy Reviews 133 (2020) 110311

[54] Ouammi A, Dagdougui H, Sacile R. Optimal control of power flows and energy [80] Zhou J, Zhang J, Cai X, Shi G, Wang J, Zang J. Design and analysis of flexible
local storages in a network of microgrids modeled as a system of systems. IEEE multi-microgrid interconnection scheme for mitigating power fluctuation and
Trans Contr Syst Technol 2015;23(1). optimizing storage capacity. Energies 2019;12(11).
[55] Brearley BJ, Prabu RR. A review on issues and approaches for microgrid [81] Yang Y, Pei W, Huo Q, Sun J, Xu F. Coordinated planning method of multiple
protection. Renew Sustain Energy Rev 2017;67:988–97. micro-grids and distribution network with flexible interconnection. Appl Energy
[56] Dewadasa M, Ghosh A, Ledwich G. Protection of microgrids using differential 2018;228:2361–74.
relays. In: Proceedings of the 2011 21st Australasian universities power [82] Abdella J, Shuaib K. Peer to peer distributed energy trading in smart grids: a
engineering conference. Brisbane, Queensland, Australia: AUPEC); 2011. p. 1–6. survey. Energies 2018;11(6).
[57] Hosseini SA, Abyaneh HA, Sadeghi SHH, Razavi F, Nasiri A. An overview of [83] Tushar W, Saha TK, Yuen C, Liddell P, Bean R, Poor HV. Peer-to-Peer energy
microgrid protection methods and the factors involved. Renew Sustain Energy trading with sustainable user participation: a game theoretic approach. Cornell
Rev 2016;64:174–86. University; 2018.
[58] Barra PHA, Coury DV, Fernandes RAS. A survey on adaptive protection of [84] Ilic D, Silva PG, Karnouskos S, Griesemer M. An energy market for trading
microgrids and distribution systems with distributed generators. Renew Sustain electricity in smart grid neighbourhoods. In: Proceedings of the 2012 6th IEEE
Energy Rev 2020;118:109524. international conference on digital ecosystems and technologies (DEST).
[59] Mirsaeidi S, Dong X, Said DM. Towards hybrid AC/DC microgrids: critical Campione d’Italia, Italy; 2012.
analysis and classification of protection strategies. Renew Sustain Energy Rev [85] Liu T, Tan X, Sun B, Wu Y, Guan X, Tsang DHK. Energy management of
2018;90:97–103. cooperative microgrids with P2P energy sharing in distribution networks. In:
[60] Bui DM, Chen S-L. Fault protection solutions appropriately proposed for Proceedings of the 2015 IEEE international conference on smart grid
ungrounded low voltage AC microgrids: review and proposals. Renew Sustain communications. Miami, Florida, USA: SmartGridComm); 2015.
Energy Rev 2017;75:1156–74. [86] Zhang C, Wu J, Zhou Y, Cheng M, Long C. Peer-to-Peer energy trading in a
[61] IEC TS 62898-1. (Microgrids - Part 1: guidelines for microgrid projects planning Microgrid. Appl Energy 2018;220:1–12.
and specification). International Electrotechnical Commission. first ed. 2017. [87] Liu N, Yu X, Wang C, Li C, Ma L, Lei J. Energy-sharing model with price-based
Switzerland; 2017. demand response for microgrids of peer-to-peer prosumers. IEEE Trans Power
[62] IEEE 2030.9-2019. IEEE recommended practice for the planning and design of the Syst 2017;32(5):3569–83.
microgrid. USA: Institute of Electrical and Electronics Engineers; 2020. [88] Tushar W, Saha TK, Yuen C, Morstyn T, McCulloch MD, Poor HV, Wood KL.
[63] Gopalan SA, Sreeram V, Iu H, Xu Z, Dong ZY, Wong KP. fault analysis of an A motivational game-theoretic approach for peer-to-peer energy trading in the
islanded multi-microgrid. In: Proceedings of the IEEE PES general meeting, San smart grid. Appl Energy 2019;243:10–20.
Diego, California, USA; 2012. p. 1–6. [89] Li Z, Kang J, Yu R, Ye D, Deng Q, Zhang Y. Consortium blockchain for secure
[64] Gopalan S, Sreeram V, Iu H, Mishra Y. A flexible protection scheme for an energy trading in industrial internet of things. IEEE Trans. Ind. Inf. 2018;14(8):
islanded multi-microgrid. In: Proceedings of the 2013 4th IEEE PES innovative 3690–700.
smart grid technologies Europe (ISGT Europe). Denmark: Lyngb; 2013. p. 1–5. [90] Luisa Di Silvestre M, Gallo P, Guerrero JM, Musca R, Sanseverino ER, Sciumè G,
[65] Abdali A, Noroozian R, Mazlumi K. Simultaneous control and protection schemes Vásquez JC, Zizzo G. Blockchain for power systems: current trends and future
for DC multi microgrids systems. Electr Power Energy Syst 2019;104:230–45. applications. Renewable and Sustainable Energy Reviews; 2019.
[66] Yaqobi MA, Matayoshi H, Danish MSS, Lotfy ME, Howlader AM, Tomonobu S. [91] Ahl A, Yarime M, Goto M, Chopra SS, Kumar NM, Tanaka K, Sagawa D. Exploring
Low-voltage solid-state DC breaker for fault protection applications in isolated DC blockchain for the energy transition: opportunities and challenges based on a case
microgrid cluster. Appl Sci 2019;9(4). study in Japan. Renew Sustain Energy Rev 2020;117:109488.
[67] Zhang F, Mu L. New protection scheme for internal fault of multi-microgrid. [92] Andoni M, Robu V, Flynn D, Abram S, Geach D, Jenkins D, McCallum P,
Protect Control Mod Power Syst 2019;4(14). Peacock A. Blockchain technology in the energy sector: a systematic review of
[68] Xu Z, Yang P, Zheng C, Zhang Y, Peng J, Zeng Z. Analysis on the organization and challenges and opportunities. Renew Sustain Energy Rev 2019;100:143–74.
Development of multi-microgrids. Renew Sustain Energy Rev 2018;81:2204–16. [93] Ahl A, Yarime M, Tanaka K, Sagawa D. Review of blockchain-based distributed
[69] Bullich-Massagué E, Díaz-González F, Aragüés-Peñalba M, Girbau-Llistuella F, energy: implications for institutional development. Renew Sustain Energy Rev
Olivella-Rosell P, Sumper A. Microgrid clustering architectures. Appl Energy 2019;107:200–11.
2018;212:340–61. [94] Mengelkamp E, Gärttner J, Rock K, Kessler S, Orsini L, Weinhardt C. Designing
[70] Kumar D, Zare F, Ghosh A. DC microgrid technology: system Architectures, AC microgrid energy markets: a case study: the Brooklyn Microgrid. Appl Energy
grid interfaces, grounding schemes, power quality, communication networks, 2018;210:870–80.
applications, and standardizations aspects. IEEE Access 2017;5. [95] Wang N, Xu W, Xu Z, Shao W. Peer-to-Peer energy trading among microgrids with
[71] Justo JJ, Mwasilu F, Lee J, Jung J-W. AC-microgrids versus DC-microgrids with multidimensional willingness. Energies 2018;11(12).
distributed energy resources: a review. Renew Sustain Energy Rev 2013;24: [96] Mihaylov M, Moffaert KV. NRGcoin: virtual currency for trading of renewable
387–405. energy in smart grids. In: Proceedings of the 11th international conference on the
[72] Planas E, Andreu J, Gárate JI, Martínez de Alegría I, Ibarra E. AC and DC European energy market (EEM14). Poland: Krakow; 2014.
technology in microgrids: a review. Renew Sustain Energy Rev 2015;43:726–49. [97] Zhang C, Wu J, Long C, Cheng M. Review of existing peer-to-peer energy trading
[73] Shahnia F, Bourbour S. A practical and intelligent technique for coupling multiple projects. Energy Procedia 2017;105:2563–8.
neighboring microgrids at the synchronization stage. Sustain Energy Grids [98] Parag Y, Sovacool BK. Electricity market design for the prosumer era. Nat Energy
Network 2017;11:13–25. 2016;1.
[74] Wu P, Huang W, Tai N, Liang S. A novel design of architecture and control for [99] Sousa T, Soares T, Pinson P, Moret F, Baroche T, Sorin E. Peer-to-peer and
multiple microgrids with hybrid AC/DC connection. Appl Energy 2018;210: community-based markets: a comprehensive review. Renew Sustain Energy Rev
1002–16. 2019;104:367–78.
[75] Majumber R. Aggregation of microgrids with DC system. Elec Power Syst Res [100] Prosumers’ role in the future energy system. Centre for Sustainable Energy
2014;108:134–43. Studies; 2018. Available online: https://fmezen.no/wp-content/uploads/2019/0
[76] Bala S, Venkataramanan G. Autonomous power electronic interfaces between 6/CenSES-position-paper-prosumer-FINAL-Language-checked.pdf. [Accessed 5
microgrids. In: Proceedings of the 2009 IEEE energy conversion congress and December 2019].
exposition (ECCE). USA: San Jose, California; 2009. p. 3006–13. [101] Souza ACZ, Castilla M. Microgrids design and implementation. Springer
[77] Goyal M, Ghosh A. Microgrids interconnection to support mutually during any International Publishing; 2019.
contingency. Sustain Energy Grids Network 2016;6:100–8. [102] Thakur S, Breslin JG. Peer to peer energy trade among microgrids using
[78] Majumber R, Bag G. Parallel operation of converter interfaced multiple blockchain based distributed coalition formation method. Technol Econ Smart
microgrids. Int J Electr Power Energy Syst 2014;55:486–96. Grids Sustain Energy 2018;3(1):1–17.
[79] Yoo H-J, Nguyen T-T, Kim H-M. Multi-frequency control in a stand-alone multi-
microgrid system using a back-to-back converter. Energies 2017;10(6).

17

You might also like